SPE 126703 Accurate and Fast Method for Predicting Actual Top-of-Cement Depths in Eccentric Wellbores

Sairam PKS, Mark Savery and, Ronnie Morgan, Halliburton

Copyright 2010, Society of Petroleum Engineers This paper was prepared for presentation at the SPE North Africa Technical Conference and Exhibition held in Cairo, Egypt, 1417 February 2010. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract It is often perceived that a cementing operation is the last step in the drilling process, but any cementer can tell you that cementing is the first step to a successful completions process. Proper displacement of mud during cementing operations is not only critical for successful zonal isolation, but also for achieving asset-management objectives for increasingly difficult-to-reach hydrocarbon reservoirs. One of the challenges in these complex wells is to provide uniform cement coverage across the entire zone of interest, especially in targeted zones where casing or liner is decentralized (off-center). In these situations, an uneven pressure differential from decentralization causes uneven flow of cement. The result is a much different top-of-cement (TOC) on the narrow side of the annulus versus the wide side of the annulus. Real-world negative effects of this phenomenon can include annular pressure buildup, the necessity of remedial squeeze work, and even well abandonment. A better understanding of fluid movement through the eccentric annuli can help predict the true TOC around decentralized casing. Therefore, a first-of-its-kind flow-predictor program was developed with the capability to accurately predict the TOC and corresponding fluid velocities for a three-component mud-spacer-cement system in eccentric annuli. It is based on a generalized mathematical model for pressure-drop prediction, developed using more than 1,700 experimental data points in physical models mimicking that of actual downhole geometries. Experimental data consisted of narrow and wide-side flow rates for 14 different fluids, within a density range of 8 to 21-lbm/gal, circulated in standard oilfield pipe and hole configurations ranging from 4.5-in. 6.5-in to 9.625-in 12.25-in. Standoffs varied between 12 and 82%. Yield point, viscosity, and shear thinning indices of these fluids varied between 0 and 50-lbf/100ft2, 1 and 157-cp, and 0.4 and 1, respectively. Various simulations are shared to reveal the properties and parameters required to help achieve uniform flow around the annulus and a balanced TOC. Introduction Operators today frequently encounter situations in which it is difficult to deploy centralization casing equipment during cementing operations. Sometimes there are mechanical constraints and torque-drag limitations that reduce the number of centralizers that can be placed on the casing. Early on, the common best practice for achieving a successful annular displacement in wellbores with low casing standoff was to apply the highest flow rate possible. While this still remains valid today, the fracture gradients in the wellbore dictate the ceiling on the flow rates and often set the boundaries on density and rheology of the annular fluids. For example, on many production liners and slim-well completions, the wellbore geometry is such that low flow rates and thin slurries are required during cementing to prevent losses. As a result, cement returns at surface or mud line do not ensure complete cement coverage around the pipe. It is clear that a better understanding of fluid movement through eccentric annuli over a wide range of casing standoff, flow rates, and fluid properties is essential to help ensure proper cement-slurry placement. Haciislamoglu and Langlinais (1990) provided a numerical solution to the equation of flow for yield power law fluids. Tehrani et al. (1992) studied the laminar displacement of fluids in annuli from both an experimental and a theoretical standpoint. Bannister and Benge (1981) studied turbulent flows from a rheometry standpoint. Li and Novotny (2006) studied the effects of density and viscosity using the Lattice Boltzmann method. Ihoyo and Azar (1981) proposed a slot-flow model for flow prediction of nonNewtonian fluids through eccentric annuli for laminar-flow regimes. Hemphill and Ravi (2005) studied the coupled effects of pipe rotation on hole cleaning and pressure drop in an eccentric annulus. More recently, Moran and Savery (2007) published the results of their experimental studies on fluid movements in eccentric annuli. They noted that future work would include creating a

SPE 126703

mathematical model to develop velocity-ratio prediction algorithms. In the following sections, the data sets and the mathematical modeling approach leading to the development of the flow-predictor program that captures the effects of pump rate, fluid rheology, standoff, hole diameter, and casing diameter are presented. It is also shown that this model was extended to two and three fluids to quickly predict the fluid positions and annular velocities during a cementing job. Methodology Large-Scale Testing Large-scale apparatus were built to conduct a complete study of steady-state response of various fluids in eccentric annuli. Experimental test data consisted of narrow-side and wide-side flow rates for 14 different fluids within a range of varying densities from 8 to 21 lbm/gal. The tests were conducted in three different model geometries at five different inner-pipe standoff values. These were pipe-in, pipe-eccentric models (Table 1), built using standard casing materials. The large geometry had an annular length of 82-in. Both the medium geometry and small geometry had annular lengths of 84in. The lengths were chosen to help ensure an L/D ratio of no less than 40 for any of the models. This was an attempt to limit the influence of any transient effects on the final result and to stabilize entrance effects. At the top of each model, two steel dividers were welded across the annular gap to split the narrow-side and wide-side flow just before exiting the model. Flow was introduced from a plenum at the bottom where flow entered the plenum from a direction perpendicular to the vertical axis. This is illustrated in Fig. 1. Fluid entered the plenum at the bottom, flowed into the vertical annulus, and exited the top where dividers separated the wide and narrow sides. Each exit led to its own collection tank atop a 1,200-lb (544-kg) rated platform scale. After testing was completed, the platform scale weight data, in conjunction with the fluid density, was used to determine the different velocities and actual flow rates. The total number of data points used in this study was 866. Mathematical Modeling To help ensure more accurate mathematical modeling and associated calculations, the rheometer readings taken for each fluid regression were best-fitted into the Generalized Herschel-Bulkley model used by Becker et al. (2003). The range of specific fluid properties of yield stress, high-shear-limiting viscosity, and shear-thinning index were 0 to 50-lbf/100ft2, 1 to 157-cp, and 0.4 to 1, respectively. Large-scale data were analysed for pressure drop across length for wide-side versus narrow-side gaps by considering a pie-slot geometry and the friction module for laminar and turbulent regimes. Mathematical modeling was carried out with the ultimate objective of modeling total pressure drop in each test section. To simulate hydraulic similarity, a flow-adjustment ratio was defined and computed. This ratio converted the experimental annular flow obtained in each segment into an equivalent flow rate in a full annulus whose outer diameter was assumed equal to the equivalent slot-flow diameter Do, eq. The equivalent flow rate (Qeq) was calculated to help ensure the average velocity in the full annulus was the same as the experimental velocity, to ensure kinetic similarity. Taking the geometry and the experimental flow rates as inputs, pressure drops were calculated independently for the narrow side and the wide side. Because of the complex nature of coupling non-Newtonian flow with non-concentric wellbore geometries, Buckinghams Pi Theorem (Murphy 1950) was deployed to develop a model of dimensionless parameters using the product-solution approach (or, in some cases, a combination of the product-solution and sum-solution approaches). These dimensionless equations using the industry accepted Similitude approach were used to model friction factors and entrance-loss coefficients for both the laminar and turbulent flow regimes (Murphy 1950). These coefficients contained a number of global constants that were optimized, along with the shear-rate coefficient for non-Newtonian fluids (Kg) and the critical Reynolds number (Recrit) to equalize the pressure drop on the narrow side and the wide side. Once the mathematical model was formulated, these global coefficients were initialized and the pressure drop was calculated in the wide side (represented by the y-coordinate) and also in the narrow side (represented by the x-coordinate). The calculations carried out for all the experimental data are presented on an XY scatter plot in Fig. 2. The global constants were simultaneously iterated until a straight-line regression fit with the minimum coefficient of variance that can be achieved on the XY scatter plot. A remarkable improvement in the coefficient of variation (COV) was seen whilst still being able to use 99.19% of the total data in the plot. The data yielded a COV of 36.9%. The mathematical model for pressure-drop calculations that included entrance losses and frictional pressure drop, in addition to velocity head, are an accurate way to predict flow regime and pressure drop in the wide and narrow sides of the annulus for any flow condition and fluid that lies in the ballpark of the fluids tested. Because all the fluids tested and the flow conditions cover the range of density and viscosity of all the typical oilfield fluids, flow rates, Reynolds numbers, and eccentricities, it can be concluded that the pressure-drop model is valid to predict the flow without further need for large-scale testing. In summary, the mathematical modeling was carried out according the procedure described (Fig. 3).

SPE 126703

Flow-Predictor Program Development As noted in the previous section, the mathematical model developed using the large-scale test data proves extremely useful to predict flow behavior in eccentric annuli. A predictor program, intended to be used for prejob design, was built on the basis of the mathematical model. Hydrostatic effects were incorporated into the flow predictor to study the behavior of annular fluids, namely spacer fluid and cement slurry. Stage-wise volumes and pump rates are inputs to the program. Quasi-static calculations for each stage can be carried out using the flow-distribution ratio. The flow-distribution ratio () is defined as the ratio of the narrow-side flow rate to the total flow rate (Eq. 1). =

Qn , or (1-) = Qtotal

Qw . (1) Qtotal

The flow-distribution ratio is iteratively calculated for each stage. Annular flow with up to three fluids can be analyzed progressively. Based on the respective volumes of the fluids in the narrow and the wide sides, a top-of-fluid calculator within the program runs in parallel to dynamically update the hydrostatic pressure as part of the total pressure-drop calculation. Eq. 2 describes the physical phenomena governing the flow of up to three fluids in an eccentric annulus.

i =1

2 2 i v w ,i i v n ,i 3 = Pfriction, w,i + Pentrance, w,i + i ghw,i + Pfriction,n,i + Pentrance,n,i + i ghn,i + .. (2) 2 2 i =1

Outputs of the non-Newtonian flow-predictor program (Fig. 4) are calculated by solving this equation. Note that entrance losses in most scenarios should be negligible for real-life wellbore scenarios. Results and Discussion The flow-predictor program can be used to analyze flow behavior in various scenarios in a single-fluid, dual-fluid, or three-fluid (e.g., mud-spacer-cement) system scenario. Some results for the single-fluid system are discussed first. Bingham plastic fluids with a PV of 20-cp were considered. At various standoffs, the effect of YP on the flow behavior is predicted in Fig. 5. The Velocity Ratio (recall from Fig. 4 that the Velocity Ratio equals VN/VW) rapidly deteriorates with increasing YP. This is an indication that the drilling mud should have the minimum YP possible considering the efficacy of hole-cleaning operations (cuttings transport and minimal solids settling) during the drilling process and, at the same time, be circulated uniformly in both the wide and the narrow annulus at the desired velocity. This becomes important when it is a priority to remove and recondition any pockets of gelled and dehydrated mud in the hole. Fig. 5 also indicates the velocity profile deteriorates the lower the standoff percentage, justifying the practice of casing centralization. Fig. 6 shows the plot of the flow-distribution ratio versus total flow rate at 80% standoff for Bingham plastic fluids with a PV of 20-cp and varying YP values. Noteworthy is the critical flow rate to initiate circulation in the narrow side of the annulus. As can be seen, the critical flow rate required increases with increasing values of YP. The effect of geometry on this phenomenon was also studied. Fig. 6 shows the same fluid (PV of 20-cp and a YP of 20lbf/100ft2) flowing through a smaller annular cross-sectional area (a 9.625-in casing inside a 10.625-in hole versus a 12.25-in hole) may require a lower circulation rate to establish flow on the narrow side of an eccentric annulus. Fig. 7 illustrates the impact of shear thinning index and o on the Velocity Ratio with 90% standoff of 12-lb/gal fluid with of 20-cp in 9.625-in x 12.25-in annulus. Fig. 8 illustrates the impact of shear thinning index and on the Velocity Ratio with 90% standoff of 12-lb/gal fluid with o of 15-lbf/100ft2 in 9.625-in x 12.25-in annulus. The results for various three-fluid cases at the end of the cement job are also discussed here. At a standoff of 40% (generally regarded as poor centralization), the effect of cement-slurry density on the final fluid positions was studied. One hundred barrels of cement slurry was placed into the annulus at a total flow rate of 5-bbl/min. As can be seen in Figs. 9 and 10 that an increase in the density of the cement slurry, from 13 to 16-lbm/gal, enhances the fluid top in the narrow side by close to 300-ft. The fluid properties associated with these cases are shown in Table 2. In this case, non-uniformities between the cement tops in the wide and narrow annulus were predicted. However, to help ensure competent zonal isolation up to the required height throughout the annulus, analysis using the predictor program would help the job designer choose the minimum volume of cement slurry to be pumped. The effect of casing standoff on the complete cementing job can also be studied using the program. An example is shown in Figs. 11 and 12 where the standoff was 40% and 50-bbl of spacer and 100-bbl of cement slurry were pumped at a 5-bbl/min annular pump rate. Figs. 13 and 14 show the results for the same fluids and operating conditions at 70% standoff. The fluid properties associated with these cases are shown in Table 3. Much better TOC values are observed with the higher standoff. Finally, using the predictor program as a simple prejob design tool, the fluid rheologies, densities, and pump rates can be

SPE 126703

optimized to achieve near-uniform TOC, as depicted in Fig. 15. The optimized slurry properties and operational parameters corresponding to this case are shown in Table 4. Figs. 16 and 17 depict the annular velocities and the stage-wise Reynolds numbers for the same. Conclusions The following conclusions are a result of this work. The key to designing a good cementing job in eccentric wellbores is understanding the effects of geometry, rheological properties, and operational parameters on fluid flow. Large-scale testing encompassing a wide variety of geometries, flow rates, and rheological parameters was carried out to understand fluid movements in eccentric annuli. The narrow- and wide-side flow rates were measured for each of these cases. Large-scale test data was extensively analyzed. Similitude modeling was implemented to fit a mathematical model that equalized the pressure drop in the narrow and the wide annulus. The mathematical model was accurate and could be used to predict the flow-distribution ratio for any user-defined case. A powerful, non-Newtonian flow-predictor program was developed based on this mathematical model. This was extended to three fluids by including hydrostatic effects into the pressure-drop calculations. The program can be used to analyze the synergistic effects of geometry, eccentricity, yield stress, high shear-rate limiting viscosity, shear-thinning index, density, and total pump rates of each of the drilling mud, spacer, and cement slurry. The program provides a guideline for the rheological design of drilling muds to help ensure proper circulation. The rheological properties and pump rates of the spacer and cement slurry can be designed according to the well conditions and the mud properties. Top of fluids, Reynolds numbers, and stage-wise velocities that are processed by the program can aid in effective cementjob design, thereby reducing the risk of costly remedial jobs caused by poorly designed fluids and/or insufficient cement coverage in the annulus. While a quick screening of many sets of rheological parameters and operational parameters can be done using this program, more value can be added to the cement-job design by performing rigorous simulations in a 3-D wellbore environment by using the properties and parameters designed by using the program. This approach is expected to save job-design time as well. Acknowledgements The authors thank the management of Halliburton for permission to publish this paper.

= annular eccentricity fraction = flow distribution ratio e = effective apparent viscosity = e / e

= viscosity = Mu_Inf = finite high-shear-limiting viscosity = fluid density e = effective shear stress directly relating to frictional pressure loss o = fluid yield stress or shear stress at near-zero shear rate ref = reference fluid shear stress, = 47.88 Pa = 1 lbf/ ft2 = viscometric shear rate e = effective fluid shear rate directly relating to frictional pressure loss aturb, bturb, cturb, dturb, eturb, fturb = turbulent flow constants alam, blam, clam, dlam, elam, = laminar flow constants Anarrow = cross-sectional area, narrow side Awide = cross-sectional area, wide side COV = coefficient of variation Di = inside diameter Do,eq = equivalent slot flow diameter fpred = predicted friction factor flam = laminar friction factor fturb = turbulent friction factor

Nomenclature

SPE 126703

He = Hedstrom number Kg = shear-rate coefficient for non-Newtonian fluids Kpred = entrance-effects coefficient for kinetic energy Apred = entrance-effects parameter Bpred = entrance-effects parameter Cpred = entrance-effects parameter g = gravitational constant hn,i = narrow-side fluid height in annulus for i fluid hw,i = narrow-side fluid height in annulus for i fluid L = length of annular flow path n = variable shear-rate exponent of a Hershel-Bulkley fluid
B

References
Bannister, C. and Benge, O. 1981. Pipe Flow Rheometry: Rheological Analysis of Turbulent Flow System Used for Cement Placement. Paper SPE 10216 presented at the Annual Technical Conference and Exhibition, San Antonio, Texas, 57 Oct. DOI: 10.2118/10216-MS. Becker, T., Morgan, R., Chin, W.., and Griffith, J. 2003. Improved Rheology Model and Hydraulics Analysis for Tomorrows Wellbore Fluid Applications. Paper SPE 82415 presented at the Production and Operations Symposium, Oklahoma City, OK. 2325 March. DOI: 10.2118/82415-MS. Haciislamoglu, M. and Langlinais, J. 1990. Non-Newtonian Flow in Eccentric Annuli. Journal of Energy Resources Technology 112 (163). Hemphill, T. and Ravi, K. 2005. Calculation of Drillpipe Rotation Effects in Axial Flow: An Engineering Approach. Paper SPE 97158 presented at the Annual Technical Conference and Exhibition, Dallas, Texas, 912 Oct. DOI: 10.2118/97158-MS. Iyoho, A. and Azar, J. 1981. An Accurate Slot-Flow Model for Non-Newtonian Fluid Flow through Eccentric Annuli. Paper SPE 9447-PA. SPE Journal. 21(5): 565572. DOI: 10.2118/9447-PA. Li, X. and Novotny R. 2006. Study on Cement Displacement by Lattice-Boltzmann Method. Paper SPE 102979 presented at the Annual Technical Conference and Exhibition, San Antonio, Texas Sept. 2427. DOI: 10.2118/102979-MS. Moran, L. and Savery, M. 2007. Fluid Movement Measurements through Eccentric Annuli: Unique Results Uncovered. Paper SPE 109563 presented at the Annual Technical Conference and Exhibition, Anaheim, California, 1114 November. DOI: 10.2118/109563-MS. Murphy, G. 1950. Similitude in Engineering. The Ronald Press Company, New York. Tehrani, A., Ferguson, J., and Bittleston, S. 1992. Laminar Displacement in Annuli: A Combined Experimental and Theoretical Study. Paper SPE 24569 presented at the Annual Technical Conference and Exhibition, Washington, D.C., 47 October. DOI: 10.2118/24569-MS.

Table 1Exact Standoff Values for Manufactured Models Large, % ( 9.625 in. 12.25 in.) 81.8 66.7 55.0 28.8 12.6 Medium, % (7 in. 8.5 in.) 82.1 67.3 54.6 30.1 13.1 Small, % (4.5 in. 6.5 in.) 82.3 66.9 55.3 28.3 12.6

SPE 126703

Table 2Fluid Properties Associated with Figs. 9 and 10 9.625 in. 12.25 in. Well Depth, ft 5,000 5,000 Standoff, % 40 40 Drilling-Mud Properties Preflush/ Spacer Properties Cement-Slurry Properties

o
lbf/100ft 10 10

cp 50 50

n 1 1

lbm/gal 12.0 12.0

o
lbf/100ft 15 15

cp 30 30

n 1 1

lbm/gal 11.0 16.0

o
lbf/100ft 45 45

cp 300 300

n
1 1

lbm/gal 13.0 16.0

Table 3Fluid Properties Associated with Figs. 11 through 14 9.625 in. 12.25 in. Well Depth, Standoff, % ft 5,000 5,000 40 70 Drilling-Mud Properties

lbf/100ft 10 10

Preflush/ Spacer Properties

cp 20 20

n 1 1

lbm/gal 12.0 12.0

lbf/100ft 15 15

Cement-Slurry Properties

cp 30 30

n .7 .7

lbm/gal 13.0 13.0

lbf/100ft 45 45

cp 300 300

n
.6 .6

lbm/gal 16.0 16.0

Table 4Fluid Properties for Figs. 15 through 17 9.625 in. 12.25 in. Well Depth, Standoff, % ft 5,000 70 Preflush pump rate Cement-slurry pump rate Drilling-Mud Properties

lbf/100ft

Preflush/ Spacer Properties

cp

n 1

lbm/gal 12.0

lbf/100ft 0.0

Cement-Slurry Properties

cp

n 1

lbm/gal 8.33

lbf/100ft 25

cp 60

n
1

lbm/gal 14

10 20 8 bbl/min 5 bbl/min

SPE 126703

Fig. 1Schematic of flow distribution in eccentric annulus (left); some of the physical models used in testing (right).

Fig. 2XY Scatter plot of pressure-drop data (859 of 866 points total are active) after global optimization.

SPE 126703

Di, Do,eq, , L, A, Qeq

Calculate Volume Averaged Shear Rate

e =

n, o, , ref, e

(D

Kg v
o ,eq

Di )

Calculate Equivalent Shear Stress as

n e n e = + ref n

Calculate Effective Viscosity Similitude Modeling Pi terms Reynolds Number Hedstrom Number

e = e e

Ree =

v(Do,eq Di ) e
2

o ( Do Di ) He = 2

flam=alam (Re)blam (He)clam fturb=aturb(Re)bturb(He)cturb+dturb(Re)eturb(He)fturb

K ent = Apred ( Ree ) pred

Pentrance

B pred

( He)

C pred

v 2 p |friction = 2 fpred Do.eq Di z

LAMINAR

v2 ent = K pred 2

No

p Pfriction = |friction L z

Ree>Recrit

Ptotal
Yes

v2 = Pfriction + Pentrance + 2

TURBULENT

Fig 3 Mathematical modeling of pressure drop.

SPE 126703

Fig. 4Flow-predictor program capabilities.

Fig. 5Effect of YP on flow distribution.

Fig. 6Effect of YP on critical flow rate.

10

SPE 126703

Fig. 7Effect of shear-thinning index with a constant viscosity, but changing yield point.

Fig. 8Effect of shear-thinning index with a constant yield point, but changing viscosity.

SPE 126703

11

Fig. 9Predicted final fluid positions for 13-lbm/gal cement slurry. Narrow-side TOC is at 4,114 ft and wide-side TOC is at 2,843 ft.

Fig. 10Predicted final fluid positions for 16-lbm/gal cement slurry. Narrow-side TOC is at 3,792 ft and wide-side TOC is at 2,972 ft.

Fig. 11Predicted final fluid positions at 40% standoff. Narrow-side TOC is at 3,709 ft and wide-side TOC = 3,006 ft.

12

SPE 126703

Fig. 12Predicted annular velocities at 40% standoff.

Fig. 13Predicted final fluid positions at 70% standoff. Narrow-side TOC is at 3,443 ft and wide-side TOC = 3,055 ft.

Fig. 14Predicted annular velocities at 70% standoff.

SPE 126703

13

Fig. 15Predicted final fluid positions in a case optimized for uniform cement-slurry distribution around the annulus. Narrow-side TOC is at 3, 224 ft and wide-side TOC = 3,196 ft.

Fig. 16Predicted annular velocities corresponding to the optimized job.

Fig. 17Stage-wise Reynolds numbers corresponding to Figs. 15 and 16.